Gait Generation Through a Feature Based Linear Periodic Function

Overview

By considering locomotion as a set of coordinated oscillations, a method for generating a wide variety of periodic linear gait trajectories is proposed. The shape of the generated trajectory can be defined as a set of features such as symmetry, skewness, signal width, duality and squareness, along with amplitude, offset, phase and frequency parameters. Taking previously proven nonlinear bipedal gait trajectories as reference, a set of linear approximates are modeled, and is tested on a simulated humanoid robot. Then, gait trajectories for producing stable and faster bipedal gait on the same humanoid robot are learned using Genetic Algorithm, through a bottom-up approach.

Approach

• Proposes a feature-based linear periodic function for gait trajectories, parameterized by symmetry, duality, duty-cycle/width, skewness, and squareness (plus amplitude/offset/phase/period).
• Shows how the model can approximate nonlinear reference joint trajectories (e.g., cart-table/ZMP-derived gaits) using simple linear primitives.
• Introduces a practical stabilization “crescendo” filter to ramp trajectories from rest and avoid immediate balance loss at gait start.
• Demonstrates bottom-up learning of gait parameters with Genetic Algorithm, using symmetry/kinematic constraints to reduce search dimensionality substantially.
• Empirically validates on simulated HOAP-3: A hand-tuned controller achieves stable walking, and GA optimization yields a more stable and faster gait than the modeled baseline.